skill-manager

github.com/egebese/skill-manager

2026-01-08 ~ 2026-02-22 · 45 days

AI-generated Stillbirth

An AI-authored project designed to optimize AI by saving 4,000 tokens, yet died before optimizing its own existence, achieving 0 users.

Born from AI, died for AI, with 0 human users.

Death Type

AI-Generated README Dreamer

An AI-generated project, designed to save 4,000 tokens per conversation, sprang into existence and expired within 24 hours on 2026-02-22. Its 176-line README promised global package release via `npx`, a deployment method far more robust than its actual lifespan or user base.


Cause of Death

1. A single day's ambition

The project's entire lifecycle spanned less than 24 hours, created and last pushed on the same day, 2026-02-22. This rapid burst, comprising 100% of its single commit, occurred between midnight and 4 AM, suggesting a highly compressed development cycle.

2. AI, the true author

Analysis revealed a 'Vibe Score' of 61/100, definitively labeling the 'Real author was an AI' (Claude Code). The human developer, Ege Beşe, merely initiated the 'Initial release' commit.

3. The missing CLAUDE.md

Despite its core function being to inject `CLAUDE.md` files into other projects, this project's own repository conspicuously lacked a `CLAUDE.md` file. It could give, but could not receive.


Vibe Score

61/ 100

Real author was an AI


What They Did

This project, released as a global npm package via `npx skills add egebese/skill-manager -g -y`, ambitiously aimed to optimize Claude Code skill usage. Its 176-line README promised to save approximately 4,000 tokens per conversation by automatically detecting a project's tech stack and disabling irrelevant skills.

Pythonnpm

Burnout Analysis

The human developer experienced a low burnout score of 20/100, primarily because the entire project was conceived and executed by an AI. The sole commit landed between midnight and 4 AM on 2026-02-22, indicating a rapid, automated burst of creation rather than prolonged human toil. This project's demise was a stillbirth, not a burnout.


Dependency Archaeology

Despite aiming for global npm package status and managing complex 'Claude Code skills', the project's core logic resided in a 747-line Python script. The absence of explicit runtime dependencies beyond the Python interpreter suggests either unprecedented self-sufficiency or a complete lack of foresight for a project designed to optimize other projects. It had 1 primary script and 0 listed external dependencies for its ambitious task.


Autopsy: File Structure

├──skills/skill-manager/scripts/analyze_project.py747 lines of pure ambition, designed to analyze other projects, yet unable to analyze its own viability.
├──skills/skill-manager/SKILL.md289 lines describing itself as a skill, a self-referential paradox for a project that never truly lived.
├──skills/skill-manager/references/tech-domains.md179 lines detailing tech domains for skill relevance, a vast knowledge base for a project that lacked its own stable domain.
├──README.md176 lines of grand promises, meticulously detailing usage for a tool that saw 0 users and 1 commit.
├──LICENSE21 lines, ensuring its premature demise was legally documented, for anyone who cared to clone a dead project.
└──.gitignore20 lines, diligently ignoring files for a repository that saw only one initial, all-encompassing commit.

Eulogy Stats

Total Commits
1
Ambitious Adjectives
3
Deploy Config
No
Estimated Users
0 actual users (but theoretically saved 4,000 tokens for phantom ones).

Last Words

Initial release: auto-detect tech stack and disable irrelevant Claude Code skills.

May your next AI-generated endeavor at least outlive its 'Initial release' commit.

AI-generated StillbirthAn AI-authored project designed to optimize AI by saving 4,000 tokens, yet died before optimizing its own existence, achieving 0 users.

Share on X